Inference Test

# all imports

import sys
import os
sys.path.insert(0, os.path.abspath('..'))
sys.path.insert(0, os.path.abspath('..\\scripts'))

from fastai.vision.widgets import *
from fastai.interpret import ClassificationInterpretation
from fastbook import *
from helpers import *
import cv2

import torch
import torchvision
from torchvision.models import resnet50, ResNet50_Weights

# get filename from models folder
learner_filename = input()
learner_filename = learner_filename.replace('.pth','')
learner_filename

'resnet_50_2023May08_0614PM'

learner_path = os.path.join(get_models_dir(), learner_filename)
learner_path

'D:\\ENGINEERING\\mezcal\\models\\resnet_50_2023May08_0614PM'

# Load fastai Learner object
data_loader = make_data_loader(get_processed_data_dir(),
                                   batch_size=64)
learn = resnet_learner(data_loader, 50)
learn.load(learner_path)

D:\ENGINEERING\mezcal\venv\lib\site-packages\torchvision\models\_utils.py:208: UserWarning: The parameter 'pretrained' is deprecated since 0.13 and will be removed in 0.15, please use 'weights' instead.
  warnings.warn(
D:\ENGINEERING\mezcal\venv\lib\site-packages\torchvision\models\_utils.py:223: UserWarning: Arguments other than a weight enum or `None` for 'weights' are deprecated since 0.13 and will be removed in 0.15. The current behavior is equivalent to passing `weights=ResNet50_Weights.IMAGENET1K_V1`. You can also use `weights=ResNet50_Weights.DEFAULT` to get the most up-to-date weights.
  warnings.warn(msg)

<fastai.learner.Learner at 0x21660d39fa0>

fastai Inference Test

Iterating throught test folder

from IPython.display import Image

test_data_dir = get_test_data_dir()

for fname in os.listdir(test_data_dir):
    image_path = os.path.join(test_data_dir, fname)    

    image_object = cv2.imread(image_path)    

    prediction,_,probs = learn.predict(image_object)    
    print(f'{max(100*probs):.2f}% {prediction} - {image_path}')
    image = Image(filename=image_path, width=224)
    display(image)    

58.46% chino_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA A_1.jpg

67.26% chino_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA A_2.jpg

94.16% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA B.jpg

65.13% tobala_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA C.jpg

97.26% tobala_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA C_2.jpg

100.00% tobala_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA C_3.jpg

99.98% tobala_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA C_4.jpg

99.75% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA D.jpg

97.10% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA D_2.jpg

92.11% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA E.jpg

99.33% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA F.jpg

99.99% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA G.jpg

99.92% espadin_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRA H.jpg

44.29% chino_impuro - D:\ENGINEERING\mezcal\data\test\MUESTRAA.jpg

interp = ClassificationInterpretation.from_learner(learn)
cm = interp.confusion_matrix()

interp.plot_confusion_matrix()

interp.plot_top_losses(9)